Word Association Norms as a Unique Supplement of Traditional Language Resources

Anna Sinopalnikova,P. Smrz
2004-05-01
Abstract:The paper deals with reuse and testing of psycholinguistic resources in linguistic studies. We report on the experimental work aiming at comparing and evaluation of linguistic information provided by Word Association Norms (WAN) with that derived from other language resources (LRs), namely corpora and wordnets. Results for 3 European languages: Russian, English and Czech are presented. Quantitative and qualitative outputs of the research show that WAN can be employed to enrich the traditional LRs and serve as a tool for their consistency checking. Introduction It is generally accepted that most today’s NLP applications need quality language resources (LRs), especially those based on real data. The well-established methods gain from large corpora, various dictionaries, lexical semantic networks, databases etc. This paper deals with another kind of LR – Word Association Norms (WAN) that has been neglected, or at least not regularly used in linguistics studies till now. Results of our research clearly show that WAN can be employed to enrich the traditional LRs and serve as a tool for their consistency checking.
What problem does this paper attempt to address?